Needle versus loop diathermy excision of the transformation zone for the treatment of cervical intraepithelial neoplasia: a randomised controlled trial

Authors


Dr W. P. Soutter, Gynaecological Oncology Centre, Hammersmith Hospital, Du Cane Road, London, UK.

Abstract

Objective  To determine whether lowwwer rates or incomplete resection of cervical intraepithelial neoplasia (CIN) may be achieved by needle excision of the transformation zone (NETZ) than with loop excision (LLETZ).

Design A prospective randomised controlled trial.

Setting A gynaecological oncology centre and a teaching hospital in West London.

Population Four hundred and four women due to receive treatment for suspected CIN.

Methods  Women were randomised to receive either LLETZ or NETZ.

Main outcome measures The study was designed to demostrate a difference in the proportion of women with clear histological margins of 82% for LLETZ compared to 94% for NETZ with 90% power at a 5% significance level, allowing for absence of CIN in the treatment specimen in 15%.

Results  Four randomised women were excluded from the analysis, as they were ineligible for the study. Three hundred and forty-seven (87%) had CIN in the treatment specimen and could be included in the analysis of excision margins. More women in the NETZ arm had clear histological margins (84.8%vs 75%, (P= 0.03). The median volume of specimens in the NETZ arm was 739 mm3 larger (P= 0.33) and they were less likely to be removed in multiple pieces (2.5%vs 29.5%, RR 0.09, 95% CI 0.04 to 0.20). Needle excision took longer to perform (median treatment time 210 vs 90 seconds, P<0.0001) and surgeons more often reported the procedure as ‘difficult’ (9.5%vs 3.0%, RR = 3.17%, 95% CI 1.33 to 7.58). No difference in peri-operative or post-operative complication rates could be demonstrated between the two groups.

Conclusion  NETZ is more likely to produce a specimen in one piece and with clear margins compared to LLETZ.

INTRODUCTION

Although no precise data are available, we estimate that more than 60,000 women are treated for cervical intraepithelial neoplasia (CIN) every year in England. This may be performed by excision or by destructive techniques. Excision is preferred as it allows histological confirmation of the diagnosis and evaluation of the completeness of excision. Loop diathermy of the transformation zone (LLETZ, LEEP) is an established method that can be performed in an outpatient setting under local anaesthetic. Although initial reports described acceptable short term failure rates between 2% and 5%,1–3 since it has become more widely used, disturbingly high failure rates of 19–45% have been reported.4–8 The LLETZ procedure is performed with a loop that is made of tungsten wire and is 1.5–2.2 cm in diameter. It is an inflexible instrument in that the amount and shape of tissue removed is predetermined by the size and shape of the loop. Often, specimens are removed in more than one piece. This can make histological assessment difficult. Furthermore, it is not possible to see the deep cutting edges of a loop leading to poor results when dealing with disease inside the cervical canal.9

Needle excision of the transformation zone (NETZ) employs a straight tungsten wire. Employing the point of this in a similar manner to a knife, the size and shape of tissue excised is determined by the configuration of the colposcopic abnormality rather than by the predefined shape of the instrument. Furthermore, the deep cutting edge is not obscured during a NETZ procedure. Observational data suggest that excision margins are more often clear of CIN following NETZ than after LLETZ.10 The current study tests this hypothesis in a randomised controlled trial.

METHODS

Eleven colposcopists recruited women into the trial from two units where both LLETZ and NETZ were available. Women were considered eligible for inclusion if they fulfilled criteria for cervical treatment for CIN and had not been treated previously. This included a cytological or colposcopic suspicion of CIN I or worse and unsatisfactory colposcopic examination in the presence of moderate or severe dyskaryosis or persistent mild dyskaryosis. Women with adenocarcinoma in situ were not included in the study. Written, informed consent was obtained from all women and local research ethics committees approved the study.

Women were randomised by telephone to the trial office where a computer generated randomisation list was kept securely. Block randomisation was used with varying sized groups. LLETZ was performed using a 1.5, 1.8 or 2.2 cm tungsten diathermy loop (Rocket, Watford, UK) according to the surgeon's preference for a specific lesion. During a LLETZ procedure, surgeons used diathermy settings according to their usual practice. Needle excision was performed with a 2 cm long tungsten wire (Rocket) using a pure coagulation setting of 35 W. The intention with both techniques was to remove the specimen in one piece if possible.

It was estimated that 340 subjects with CIN should be randomised equally to each arm to detect a difference in the proportion of women with all clear margins of 82% for LLETZ compared with 94% for NETZ with 90% power at P < 0.05. There is substantial variation in the rate of complete excision reported in the literature ranging from 51% to 88.5%.11,12 The figure of 82% was chosen because, although our own retrospective study showed a 64.5% rate,10 the rate improved to about 82% when everyone became aware of the problem. Our retrospective study10 showed a 92% rate for NETZ, including our very first efforts, so it seemed reasonable to expect a small improvement on that with experience. Four hundred and four women were randomised as it was estimated that approximately 15% of women would be found to have no identifiable CIN in the treatment specimen.

Women were randomised to the study between April 1998 and March 2001. Patients were not informed of the type of treatment they received. Following treatment, the duration of treatment and any peri-operative complications were recorded. Operative pain was documented using a linear scale. Surgeons made a subjective assessment of the complexity of the procedure as having no, slight, moderate or severe difficulty. In addition, surgeons made a subjective assessment of the extent to which bleeding interfered with the procedure.

Histological analysis was performed by the routine pathological service in each hospital. Histopathologists were unaware of the type of treatment patients received. Specimens were measured and the dimensions recorded. The presence and extent of any abnormal lesion was reported in addition to the completeness of excision.

Follow up appointments were organised at 6 and 12 months. Late complications were recorded at these visits. Post-operative haemorrhage was defined as the need for admission, transfusion or suturing for post-operative bleeding. Cervical stenosis was defined as the inability to insert a cotton Q-tip into the cervical os. At each follow up visit, patients were assessed for residual or recurrent cervical dysplasia by cytology and colposcopically directed biopsies where appropriate. When cytological and colposcopic examination revealed no abnormality, women were discharged to the care of their general practitioner with a recommendation for annual smears.

An intention-to-treat approach was adopted for the analysis of outcome measures. We realised that the primary endpoint would have to be restricted to those women with CIN in the excisional specimen and adjusted the power calculation accordingly. A power analysis was performed based on data from an observational study10 using a corrected χ2 and Fisher's Exact Tests. Continuous comparative data not consistent with a normal distribution were analysed using the Mann–Whitney U test. Dichotomous data were analysed using the χ2 test and a Yates' correction and Fisher's Exact Test were used when expected or observed values were less than five.

RESULTS

Of 404 women recruited to the study, four were excluded. One was not suitable for excisional treatment due to extension to the vagina; one patient fainted during the administration of local anaesthetic; one patient was subsequently found to be ineligible due to a previous treatment to her cervix; and one patient changed her mind and requested LASER ablation outside the study (Fig. 1). Two women did not receive the treatment to which they were randomised but were analysed according to an intention-to-treat approach. One of these women from the NETZ group received a LLETZ after a failed needle excision and one woman in the LLETZ group received the wrong treatment mistakenly (Fig. 1). Three hundred and forty-seven (87%) of the original 400 women had CIN in their final specimen and were available for analysis of excision margins (Fig. 1). Forty-eight women (12%) were lost to follow up, the numbers being similar in each group (22 vs 26, Fig. 1).

Figure 1.

Flow diagram of trial.

Follow up data were available from 325 (92%) of the 347 women with CIN in their excised specimens (Fig. 1). The distribution of age, parity, pretreatment cytology, pretreatment colposcopy, position of the squamo-columnar junction (SCJ) and pretreatment histology were similar in the two groups (Table 1). Information about late complications was available for 352 (88%) of the original 400 women who received treatment.

Table 1.  Baseline characteristics of 400 randomised women. Values are given as mean [SD] or n (%).
 LLETZ n= 200NETZ n= 200
Age (years)32.4 [9.5]32.8 [9.4]
Parous80 (40.0)84 (42.0)
Treatment under local anaesthetic196 (98.0)198 (99.0)
 
Worst pretreatment smear
No pretreatment smear1 (0.5)1 (0.5)
Unsatisfactory2 (1.0)6 (3.0)
No abnormality seen6 (3.0)2 (1.0)
Non-specific inflammation0 (0.0)3 (1.5)
Borderline change12 (6.0)17 (8.5)
Warty change1 (0.5)1 (0.5)
Mild dyskaryosis66 (33.0)72 (36.0)
Moderate dyskaryosis73 (36.5)52 (26.0)
Severe dyskaryosis39 (19.5)44 (22.0)
Suspected glandular abnormality0 (0.0)2 (1.0)
 
Pretreatment colposcopy
Not recorded1 (0.5)1 (0.5)
No abnormality seen8 (4.0)6 (3.0)
Metaplasia2 (1.0)0 (0.0)
Flat warty change8 (4.0)4 (2.0)
Condylomatous warts0 (0.0)0 (0.0)
CIN I71 (35.5)63 (31.5)
CIN II64 (32.0)86 (43.0)
CIN III42 (21.0)36 (18.0)
Suspected microinvasion1 (0.5)2 (1.0)
Indecisive / unsatisfactory1 (0.5)2 (1.0)
Other2 (1.0)0 (0.0)
 
Position of SCJ
Not recorded4 (2.0)4 (2.0)
On the ectocervix100 (50.0)110 (55.0)
Seen easily in the canal80 (40.0)73 (36.5)
Seen with difficulty in the canal8 (4.0)6 (3.0)
Hidden in the canal8 (4.0)7 (3.5)
 
Pretreatment histology
No pretreatment histology20 (10.0)13 (6.5)
No abnormality2 (1.0)0 (0.0)
Unsatisfactory biopsy1 (0.5)5 (2.5)
Metaplasia only0 (0.0)1 (0.5)
Non-specific inflammation4 (2.0)4 (2.0)
Warty change6 (3.0)5 (2.5)
CIN I72 (36.0)77 (38.5)
CIN II63 (31.5)61 (30.5)
CIN III32 (16.0)34 (17.0)
Adenocarcinoma in situ0 (0.0)0 (0.0)
Invasion0 (0.0)0 (0.0)

The treatments were performed by 12 surgeons. One surgeon treated 206 women, two others treated 47 and 57 women, respectively, three treated 20–30 women each and the remainder treated fewer than 12 patients each. There was no significant difference in the distribution of the two treatments performed between the surgeons (χ2= 6.3, degrees of freedom = 11, P= 0.91, Fisher–Freeman–Halton Exact Test). Nine of the surgeons had no prior experience of NETZ. Needle excision took a median 2 minutes longer to perform than LLETZ (210 vs 90 seconds, P < 0.0001, Table 2). Surgeons reported NETZ as being moderately or severely difficult more often than LLETZ (Fisher's Exact Test, two sided P= 0.012, Table 2) and peri-operative blood loss interfered with treatment more often in the needle excision group (Fisher's Exact Test, two sided P= 0.006, Table 2). There were three peri-operative complications in the NETZ group. All were associated with bleeding that required a vaginal pack. No patients in either arm required a cervical suture or blood transfusion at the time of treatment and no patients had a cervical perforation or diathermy burn to the vagina. A similar proportion of women in each arm reported no peri-operative pain (Table 2). All but four were treated under local anaesthesia with superficial injection of 3% prilocaine and 0.03 iu/mL octapressin. Two from each group were treated under general anaesthesia.

Table 2.  Outcome of treatment. Values are given as median [interquartile range] or n (%).
 LLETZNETZ
  • *

    P < 0.0001 (Mann–Whitney U test).

  • **

    Data fields were not completed for seven women in the LLETZ arm and six in the NETZ.

  • Fisher's Exact test two sided P= 0.012.

  • ††

    Fisher's Exact test two sided P= 0.006.

  • ns

    Not statistically significant.

  • nsf

    Not statistically significant using Fisher's Exact test.

During treatmentn= 200n= 200
Duration of treatment (seconds)*90 [60–120]210 [180–300]
No peri-operative painns178 (89.0)174 (87.0)
Moderate to severe difficulty of procedure6 (3.0)19 (9.5)
Blood loss interfered with treatment††7 (3.5)22 (11.0)
Peri-operative complicationsnsf0 (0.0)3 (1.5)
 
Follow upn= 174n= 178
Antibiotics for suspected genital infectionns25 (14.4)23 (12.9)
Hospital admission for bleedingnsf5 (2.9)8 (4.5)
Blood transfusionnsf1 (0.6)2 (1.1)
Required cervical suture to control bleedingnsf2 (1.1)0 (0.0)
 
Assessment of stenosisn= 167n= 172
In ability to insert Q-tip through cervical osnsf13** (7.8)23** (13.4)

Of the 352 women who had at least one follow up appointment, there was no difference between the two arms in the number of women who required antibiotics for a genital infection (Table 2). The proportion of women who required a hospital admission, a blood transfusion or a cervical suture for secondary haemorrhage was similar in the two arms (Table 2). Nor was there any demonstrable difference in the rate of cervical stenosis between the two groups (Table 2).

In 5 (2.5%) of women in the NETZ group compared with 59 (29.5%) of women in the LLETZ group, the excision biopsy was removed in more than one piece (RR 0.09, 95% CI 0.04 to 0.20, Table 3). The median volume of the specimens was greater in the NETZ arm (median volume 3143 mm3vs 2404 mm3, P= 0.033) (Table 3). This was due to an increase in the length of the cone (median length 20 vs 14 mm, P < 0.0001). The transverse and vertical diameters were of similar dimensions (Table 3).

Table 3.  Histopathology results. Values are given as median [interquartile range] or n (%).
 LLETZ (n= 200)NETZ (n= 200)
  • *

    RR 0.09 (95% CI 0.04 to 0.20), χ2= 29.43, P < 0.0001.

  • **

    P= <0.0001 (Mann–Whitney U test).

  • P= 0.033 (Mann–Whitney U test).

  • ††

    Fisher's Exact test two sided P= 0.031.

  • ns

    Not statistically significant.

Macroscopic specimen
Specimen in two or more pieces*59 (29.5)5 (2.5)
Cone length (mm)**14 [10–19]20 [13–25]
Transverse diameter of cone (mm)ns20 [16–25]20 [15–23]
Vertical diameter of cone (mm)ns18 [15–20]17 [14–20]
Volume of cone (mm3)2404 [1230–3980]3143 [1430–5070]
 
Final histology
Unknown0 (0.0)1 (0.5)
No abnormality7 (3.5)8 (4.0)
Metaplasia only2 (1.0)3 (1.5)
Non-specific inflammation5 (2.5)3 (1.5)
Warty change6 (3.0)12 (6.0)
CIN I61 (30.5)49 (25.5)
CIN II50 (25.0)57 (28.5)
CIN III65 (32.5)65 (32.5)
Adenocarcinoma in situ0 (0.0)1 (0.5)
Micro-invasion3 (1.5)0 (0.0)
Frank invasion1 (0.5)0 (0.0)
Invasive adenocarcinoma0 (0.0)1 (0.5)
 
Excision margins clear of CIN (in those who had CIN in specimen)n= 176n= 171
Endocervicalns158 (90.0)161 (94.2)
Ectocervicalns149 (84.7)153 (89.5)
Stromalns174 (98.9)171 (100.0)
All margins clear††132 (75.0)145 (84.8)

The histological diagnoses of the final specimens were similar in the two arms (Table 3). Although 110 had only CIN I in the final specimen, 24 of these had CIN II–III in the pretreatment biopsy. Taking the results of both the punch and treatment biopsies into consideration, 79 (19.8%) had CIN I and 14 (3.5%) had no biopsy showing CIN.

More women in the NETZ group had completely clear margins histologically than in the LLETZ group (Fisher's Exact Test, two sided P= 0.031, Table 3). At least one margin was uncertain in 17 (9.7%) of the LLETZ biopsies compared with only 5 (2.9%) of the NETZ specimens (RR 0.30, 95% CI 0.12 to 0.77). Potential bias in the assessment of completeness of excision introduced by the different surgeons was investigated with a bias assessment plot of log relative risk against the standard error.13 The intercept was −0.27 (approximate 95% CI −0.91 to 0.37, P= 0.34), indicating that there was no evidence of bias introduced by the surgeons. There was no difference in the cumulative risk of developing a recurrent or residual high grade intraepithelial lesion between the two groups at follow up (log-rank χ2= 0.127, P= 0.7218) but the study would have needed over 1200 subjects to assess recurrence rates reliably.

DISCUSSION

These data demonstrate that the excision specimen is deeper and of greater volume following a NETZ than after LLETZ. This results in a higher proportion of clear margins compared with traditional LLETZ procedures. However, needle excision takes slightly longer to perform and surgeons more often report NETZ as being difficult.

This study was not large enough to show any difference in the proportion of treatment failures between the two groups and was designed to demonstrate a difference only in involved histological margins. As there is a 2.6- to 8-fold increase in treatment failures associated with involved margins following excisional treatment of the cervix,8,11,12,14–16 the results support the hypothesis that NETZ excisions may be associated with fewer treatment failures. The results are consistent with those of an observational study in which excision margins were unequivocally complete in 64.5% of patients who had LLETZ compared with 91.8% of those who had a needle excision.10

The results for LLETZ in the current study were better than in our earlier series but it should be noted that the 14 mm median length of the LLETZ specimens in this study is substantially greater than in our previous series and much greater than is usually described. One another, previous publication refers to NETZ.17 This describes an observational study of 58 women treated with a similar instrument during a 30 month period in 1990–1992 and followed for 5 years. The mean length of the biopsies was 19.8 mm and 82.8% had completely free margins. The 5 year disease-free rate was 94.8%.

The median depth of excision was greater in the NETZ arm. However, there was no statistically significant difference in the proportion of women with involved endocervical edges. The study was designed to demonstrate differences in all margins being free of disease and the lack of a statistically significant difference in the individual results for endocervical and ectocervical clearance reflects a lack of power for these subanalyses.

The reason for an improved proportion of clear resection margins in the needle excision group may be related to a number of factors. If the depth of excision was the only reason for improved resection margins, then the endocervical borders of the histological specimen would have accounted for all the differences between the two groups. An improved rate of excision at the ectocervical margins might be because NETZ can outline the ectocervical margins of a large lesion more readily than can LLETZ and the excision can be tailored more closely to the distribution of the lesion.

The number of pieces in the final histopathological specimen must also contribute to the assessment of the margins because a single specimen is easier to assess for margin involvement than multiple pieces of tissue. It is possible that the larger number of LLETZ specimens removed in multiple fragments increased the number of apparent incomplete excisions—a supposition supported by the higher rate of LLETZ specimens in which the status of at least one of the margins was uncertain. If so, the assumption that a reduced rate of involved margins in the NETZ group would result in fewer treatment failures may not be true. Even if that were the case, knowledge of margin involvement has other important clinical roles including the accurate counselling of patients concerning their individual risk of recurrence or residual disease. A number of authors have advocated increased colposcopic surveillance for patients with involved margins.8,11,12,14–16 Needle excision would result in fewer women having unnecessary colposcopic follow up because the pathological margins of the specimen could not be confirmed as being unequivocally uninvolved.

There were no differences in complication rates demonstrated between the two groups in this study. However, surgeons more often reported that the NETZ procedure was difficult and were more often troubled by bleeding during needle excision. This was probably related at least in part to their relative lack of experience of this new technique. The trial did not look at complications in subsequent pregnancies. Excisional treatment of the cervix may increase the risk of preterm labour if the depth of excision exceeds 1 cm.18 In this study, although the mean depth of excision was greater in the needle diathermy arm, this seems to reflects the greater flexibility of NETZ to tailor the depth of excision according to the requirements of a specific lesion.

In summary, NETZ is more likely than LLETZ to result in a specimen that is removed in one piece and has unequivocally clear margins. The shape, size and depth of a NETZ can be tailored precisely to the lesion, and for this reason, it is especially valuable in treating women who have large lesions or whose disease involves the cervical canal. However, it does take longer to learn.

Conflict of Interests

There are no conflicts of interests.

Accepted 3 March 2004

Ancillary